Brain arteriovenous malformation (bAVM) patients are at risk of intracranial hemorrhage (ICH). Roughly 20% of patients are not offered treatment because of excessive risk of invasive therapy, and treatment of unruptured lesions has become controversial. There are no specific medical therapies to treat bAVMs. We have developed novel adult-onset mouse models using homozygous conditional deletion of Activin-like kinase (Alk1) and Endoglin (Eng), the causative genes for a familial form of bAVM (Hereditary Hemorrhagic Telangiectasia). The models have phenotypes mimicking key aspects of human bAVM. Aims 1 and 2 will address how loss of function of these two genes results in essentially the same human bAVM phenotype, which is not currently understood, while Aim 3 will develop much-needed therapy for bAVM. Aim 1 examines how loss of Alk1 or Eng gene function alters signaling that is crucial for structural integrity of the vascular wall. For Aim 1a (in vitro) and im 1b (in vivo), the overarching hypothesis is that: (1) ALK1 or ENG deletion with VEGF stimulation leads to reduced expression of the Notch ligand, Delta-like ligand-4 (DLL4), in brain microvascular endothelial cells; (2) decreased DLL4 results in deficient platelet-derived growth factor-B (PDGFB) signaling; (3) consequently, mural cell recruitment is impaired. Aim 2 focuses on the cellular loci where the loss of gene function acts. We hypothesize that Alk1 or Eng-deficient bone marrow (BM)-derived endothelial cells (EC) and/or macrophages (Mo) are sufficient to induce dysplasia (bAVM phenotype). We will test whether BM-derived EC (Aim 2a) or BM-derived Mo (Aim 2b) are sufficient to induce dysplasia. In Aim 2c, we will test for synergism between the two cells types, and in Aim 2d, we will use cell-specific overexpression of Notch signaling (Notch1 in Mo or Dll4 in EC) to rescue the effects of Alk1 or Eng deletion. Aim 3 will provide pre-clinical data on one promising drug that was selected, in part, based on our hypotheses in Aim 1. Our data suggest that thalidomide decreases the severity of dysplasia and stabilizes the vascular wall. Because of thalidomide's well-known adverse effect profile, we will test lenalidomide, a new, less toxic analogue. In Aim 3a, we hypothesize that lenalidomide is superior to thalidomide. In Aim 3b, we will determine efficacy of lenalidomide by constructing dose-inhibitory curves in our Alk1 deletion model. In Aim 3c, to insure broader applicability, we will test lenalidomide in our Eng deletion model using a near-maximally effective dose. These pre-clinical data can be used to support early-phase clinical trials. The proposed work will underpin continued efforts to elucidate the mechanisms of bAVM pathogenesis and improve care for bAVM patients.